AWS EKS - cluster no nodes availabe to schedule pods

1
votes

I am new to AWS-EKS and am trying to create cluster and attack worker nodes and i have followed AWS documentation https://docs.aws.amazon.com/eks/latest/userguide/getting-started-console.html and using cloudformation for creating VPC with both private and public subnets and launched cluster using both private and public subnets then configured kubectl,AWS CLI and Installing aws-iam-authenticator as guided in documentation and Launched Amazon EKS Linux Worker Nodes using cloudformation yaml format and ran all commands to enable worker nodes to join your cluster but when i run kubectl get nodes i get result as No resource found.

i have run below commands to troubleshoot:

kubectl get pods following is the response

kubectl get events following is the response

i have installed kubectl on linux instances and worker nodes as well in linux instances. Please help me to fix this issue.

1
amazon-web-serviceskuberneteskubernetes-podamazon-eks
The machine you're using to run kubectl commands should be configured with the same IAM creds that created the EKS cluster. Have you checked for that? - Mrudav Shukla
Hi Mrudav, Thank you for response. Is that a access keyid and secret access key in aws CLI? I have admin privilege in AWS Console and corresponding access key id and secret access key that is wat i used in aws cli. - heena r
Ok. And the node groups that you added for this EKS cluster are up and running? - Mrudav Shukla
Ya they are running - heena r
This always happens due to some network misconfiguration. Make sure you have used the correct AMI for that particular k8s version. Also check that nodegroup has proper iam roles and the associated security groups are allowed to connect to the control plane. - rohanmehto2

1 Answers

1
votes

On top of my head 3 things can cause this:

  1. aws-auth might not be setup properly. Make sure kubectl get cm -n kube-system aws-auth -oyaml exists and contains the right rolearn given by the cloudformation output (node group role arn)
  2. The ControlPlane security group might not allow the subnets of your nodes.
  3. The ClusterName does not match the actual EKS ClusterName